1 Aggregation Latency-Energy Tradeoff in Wireless Sensor Networks with Successive Interference Cancellation Hongxing Li, Chuan Wu, Dongxiao Yu, Qiang-Sheng Hua and Francis C.M. Lau Department of Computer Science, The University of Hong Kong, Hong Kong Email: {hxli, cwu, dxyu, qshua, fcmlau}@cs.hku.hk Abstract—Minimizing latency and energy consumption is the prime objective of the design of data aggregation in battery-powered wireless networks. A tradeoff exists between the aggregation latency and the energy consumption, which has been widely studied under the protocol interference model. There has been however no investigation of the tradeoff under the physical interference model which is known to capture more accurately the characteristics of wireless interferences. When coupled with the technique of successive interference cancellation, by which a receiver may recover signals from multiple simultaneous senders, the model can lead to much reduced latency but increased energy usage. In this paper, we investigate the latency-energy tradeoff for data aggregation in wireless sensor networks under the physical interference model and using successive interference cancellation. We present theoretical lower bounds on both latency and energy as well as their tradeoff, and give an efficient approximation algorithm that can achieve the asymptotical optimum in both aggregation latency and latency-energy tradeoff. We show that our algorithm can significantly reduce the aggregation latency, for which the energy consumption is kept at its lowest possible level. Index Terms—Data aggregation, Latency-energy tradeoff, Wireless sensor network, Successive interference cancellation. ✦ 1 I NTRODUCTION Wireless sensor networks have been extensively exploited for many environment monitoring applications in recent years. One of the core functions in these networks is data aggregation, which is to collect data from the wireless sensor nodes to deliver to a sink node. Typically, data aggregation is initiated by the sink using some SQL-like queries, such as “to find the highest temperature in the region”. Messages generated at individual sensors carrying temperature data, are first aggregated and processed at some relay sensors, e.g., to derive the local maximum temperature; the locally processed results are further aggregated, and so on, until the final result reaches the sink. Besides the max function, other functions such as min, sum, count, and average can all be effectively implemented using data aggregation. As the sensed data typically has a limited duration of validity, a fundamental requirement is that the total ag- gregation time, measured in time units and also referred to as the aggregation latency, must be minimized [1]–[3]. Additionally, the sensor nodes have to observe the hard constraint imposed by battery power and must strive for low energy consumption in each run of the data aggregation. Obviously, there exists some kind of tradeoff between aggregation latency and energy consumption (the latency- energy tradeoff) in wireless sensor data aggregation [4]–[6]. There have been some efforts to derive latency-energy tradeoff theoretically [7] as well as practical algorithms [4]– [6], which are all based on the protocol interference model (or equivalently the pair-wise interference model). Under the protocol interference model, the transmission range and interference range of a node are simplified to two disks with radii r t and r i (r i ≥ r t ), respectively. A transmission is successful if and only if the receiver lies within the transmis- sion range of the sender and outside the interference range of any other concurrent sender. There has been however no prior study that is based on the physical interference model (or the cumulative interference model) which has been shown to be able to more accurately characterize the wireless interferences than the protocol interference model [8]–[10]. Designs based on the physical interference model can lead to increased network capacity. Under the physical interference model, the cumulative interference from all concurrent transmissions, e.g. the ∑ ej ∈Λi P j /d α ji part in Eqn. (1), is taken into consideration at each receiver. A transmission along link e i is successful if the Signal-to- Interference-plus-Noise-Ratio (SINR) at its receiver is above a certain threshold: Pi /d α ii N0 + ∑ e j ∈Λ i Pj /d α ji ≥ β. (1) Here Λ i denotes the set of links that transmit simultaneously with e i . P i and P j denote the transmission powers at the transmitter of link e i and that of link e j , respectively. d ii (d ji ) is the distance between the transmitter of link e i (e j ) and the receiver of link e i . Fig. 1 explains these distances graphically. α is the path loss ratio which has a typical value of between 2 to 6. N 0 is the ambient noise power. β is a positive constant as the SINR threshold for a successful transmission [3], [11]. ii ji jj i j Fig. 1: An illustration of distances with two transmission links: e i and e j . With the physical interference model, a receiver can only successfully recover one signal from one sender in each